Knowledge Prompting in Pre-trained Language Model for Natural Language Understanding
This work addresses the challenge of efficiently integrating knowledge into PLMs for natural language understanding, offering a flexible approach that avoids structural changes and irrelevant knowledge, with incremental improvements over existing methods.
The paper tackles the problem of incorporating factual knowledge into pre-trained language models without modifying their internal structures, proposing a knowledge prompting framework (KP-PLM) that converts knowledge sub-graphs into natural language prompts and uses self-supervised tasks, achieving superior performance on multiple NLU tasks in full-resource and low-resource settings.
Knowledge-enhanced Pre-trained Language Model (PLM) has recently received significant attention, which aims to incorporate factual knowledge into PLMs. However, most existing methods modify the internal structures of fixed types of PLMs by stacking complicated modules, and introduce redundant and irrelevant factual knowledge from knowledge bases (KBs). In this paper, to address these problems, we introduce a seminal knowledge prompting paradigm and further propose a knowledge-prompting-based PLM framework KP-PLM. This framework can be flexibly combined with existing mainstream PLMs. Specifically, we first construct a knowledge sub-graph from KBs for each context. Then we design multiple continuous prompts rules and transform the knowledge sub-graph into natural language prompts. To further leverage the factual knowledge from these prompts, we propose two novel knowledge-aware self-supervised tasks including prompt relevance inspection and masked prompt modeling. Extensive experiments on multiple natural language understanding (NLU) tasks show the superiority of KP-PLM over other state-of-the-art methods in both full-resource and low-resource settings.